src/Gargantext/Core/Text/Corpus/Parsers/Gitlab.hs

   1 module Gargantext.Core.Text.Corpus.Parsers.Gitlab (
   2   Issue(..), gitlabIssue2hyperdataDocument, readFile_Issues, readFile_IssuesAsDocs
   3 ) where
   4
   5 import Data.Aeson
   6 import Data.Time
   7 import qualified Data.Text            as DT
   8 import qualified Data.ByteString.Lazy as DBL
   9 import System.FilePath (FilePath)
  10
  11 import Gargantext.Prelude
  12 import Gargantext.Core (Lang(..))
  13 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
  14
  15 data Issue = Issue { _issue_id      :: !Int
  16                    , _issue_title   :: !DT.Text
  17                    , _issue_content :: !DT.Text
  18                    , _issue_created :: !LocalTime
  19                    , _issue_closed  :: !(Maybe UTCTime)
  20                    }
  21   deriving (Show)
  22
  23 instance FromJSON Issue where
  24   parseJSON = withObject "Issue" $ \v -> Issue
  25     <$> v .: "c0" -- id
  26     <*> v .: "c1" -- title
  27     <*> v .: "c2" -- content
  28     <*> v .: "c3" -- creation time
  29     <*> v .:? "c4" -- close time
  30
  31 gitlabIssue2hyperdataDocument :: Issue -> HyperdataDocument
  32 gitlabIssue2hyperdataDocument issue = HyperdataDocument
  33     { _hd_bdd = Nothing
  34     , _hd_doi = Nothing
  35     , _hd_url = Nothing
  36     , _hd_uniqId = Nothing
  37     , _hd_uniqIdBdd = Nothing
  38     , _hd_page = Nothing
  39     , _hd_title = Just (_issue_title issue)
  40     , _hd_authors = Nothing
  41     , _hd_institutes = Nothing
  42     , _hd_source = Nothing
  43     , _hd_abstract = Just (_issue_content issue)
  44     , _hd_publication_date = Just $ DT.pack $ show date
  45     , _hd_publication_year = Just $ fromIntegral year
  46     , _hd_publication_month = Just month
  47     , _hd_publication_day = Just day
  48     , _hd_publication_hour = Just (todHour tod)
  49     , _hd_publication_minute = Just (todMin tod)
  50     , _hd_publication_second = Just (round $ todSec tod)
  51     , _hd_language_iso2 = Just $ (DT.pack . show) lang
  52     }
  53   where lang = EN
  54         date = _issue_created issue
  55         (year, month, day) = toGregorian $ localDay date
  56         tod = localTimeOfDay date
  57
  58 readFile_Issues :: FilePath -> IO [Issue]
  59 readFile_Issues fp = do
  60   raw <- DBL.readFile fp
  61   let mayIssues = decode raw
  62   case mayIssues of
  63     Just is -> pure is
  64     Nothing -> pure []
  65
  66 readFile_IssuesAsDocs :: FilePath -> IO [HyperdataDocument]
  67 readFile_IssuesAsDocs = fmap (fmap gitlabIssue2hyperdataDocument) . readFile_Issues